Machine learning phenotyping and GWAS reveal genetic basis of Cd tolerance and absorption in jute

Zemao Yang,Alei Li,Jiquan Chen,Zhigang Dai,Jianguang Su,Canhui Deng,Gaoao Ye,Chaohua Cheng,Qing Tang,Xiaoyu Zhang,Ying Xu,Xiaojun Chen,Bibao Wu,Zhihai Zhang,Xuying Zheng,Lu Yang,Liang Xiao
DOI: https://doi.org/10.1016/j.envpol.2024.124918
2024-09-10
Abstract:Cadmium (Cd) is a dangerous environmental contaminant. Jute (Corchorus sp.) is an important natural fiber crop with strong absorption and excellent adaptability to metal-stressed environments, used in the phytoextraction of heavy metals. Understanding the genetic and molecular mechanisms underlying Cd tolerance and accumulation in plants is essential for efficient phytoremediation strategies and breeding novel Cd-tolerant cultivars. Here, machine learning (ML) and hyperspectral imaging (HSI) combining genome-wide association studies (GWAS) and RNA-seq reveal the genetic basis of Cd resistance and absorption in jute. ML needs a small number of plant phenotypes for training and can complete the plant phenotyping of large-scale populations with efficiency and accuracy greater than 90%. In particular, a candidate gene for Cd resistance (COS02g_02406) and a candidate gene (COS06g_03984) associated with Cd absorption are identified in isoflavonoid biosynthesis and ethylene response signaling pathways. COS02g_02406 may enable plants to cope with metal stress by regulating isoflavonoid biosynthesis involved in antioxidant defense and metal chelation. COS06g_03984 promotes the binding of Cd2+ to ETR/ERS, resulting in Cd absorption and tolerance. The results confirm the feasibility of high-throughput phenotyping for studying plant Cd tolerance by combining HSI and ML approaches, facilitating future molecular breeding.
What problem does this paper attempt to address?